Weighted Co-Occurrence Bio-Term Graph for Unsupervised Word Sense Disambiguation in the Biomedical Domain

Word Sense Disambiguation (WSD) is a significant and challenging task for text understanding processing. This paper presents an unsupervised approach based on Weighted Co-occurrence bio-Term Graph (WCOTG) performing WSD in the biomedical domain. The graph automatically created from terms that are extracted corpus of downloaded scientific abstracts. Two kinds weights introduced links built bio-term taken as important factors process disambiguation. modified Personalised PageRank (PPR) algorithm used WSD. When evaluated NLM-WSD MSH-WSD test datasets, acronym set, method outperforms widely ones addressing same problem, average result almost equal to BlueBERT_LE-based method. In contrast, our has no additional enhancement or training BERT-based models. Comparative experiments validate positive effect links’ weight disambiguation efficiency. Last, statistical relation among system accuracy, numbers medical abstracts corpus, corresponding suggest excellent minimum scale, when resources limited.